Comuptional reasoning and learning for smart manufacturing under realistic conditions

Smart manufacturing has increasingly become a prominent research topic across the academia and industry during recent years. However, in the real world, most factories' conditions are normally insufficient for implementing full scale smart manufacturing. In order to increase smartness for manufacturing systems under different situations, one may observe that computational reasoning and learning, including latest machine learning methodology and traditional rule based systems, are able to offer potential powerful theoretical foundations as well as technical tools for enabling such smarter systems. Unfortunately, few studies exist to propose a practical yet systematic procedure that implements computational reason and learning to implement smartness for typical manufacturing systems. This paper proposes a general computational reasoning and learning framework to describe the key functions in smart manufacturing under the "three Fs in one" system framework, namely a system of interconnected data, integrated automation, and intelligent information. Among three Fs, the intelligent information plays the most important role in working towards smartness by connecting the other two Fs. Furthermore, to achieve it, a learning enabled comprehensive multi-agent decision model is developed. In particular, we first design a computational learning based architecture for analyzing support information for manufacturing processes. Then we provide an optimization architecture that enables realtime learning for a manufacturing process. At last, we employ a rule based learning system to integrate these two architectures to facilitate self-evolution of the manufacturing system. The advantages of our procedure include adaptive responses to dynamic environment, efficient computations, and abilities to fulfill complex manufacturing processes, which are demonstrated by a specific modeling procedure.